Background Structural lesions are visible on MRI T1-weighted images of the sacroiliac (SI) joints. However, radiographs are currently the established method to detect structural lesions (SL).
Objectives To compare the performance of MRI to radiographs in detecting SL in the SI-joints.
Methods The SPondyloArthritis Caught Early (SPACE)-cohort includes patients (pts) with back pain (≥3 months, ≤2 years, onset <45 years) from the 5 participating centers. Pts underwent MRI and radiographs of the SI-joints (MRI-SI and X-SI, resp). MRI-SI (T1-weighted images; STIR viewed simultaneously) and X-SI were scored by 3 well-calibrated readers independently on the presence of erosions, sclerosis and ankylosis. Erosions and sclerosis were defined according to the MORPHO definition1 (≥1 lesion on ≥2 consecutive slices or ≥2 lesion on 1 slice). Ankylosis was defined as ≥1 lesion on ≥1 slice. Lesions were considered present if 2/3 readers agreed. MRI was considered positive for SL (MRI-SI-s+) if ≥4 or ≥5 SL (any type) were present. Comparisons between MRI-SI and X-SI using crosstabs (Kappa) were done for erosions, sclerosis, ankylosis and for MRI-SI-s+ versus modified New York (mNY) score2.
Results Pts with complete MRI-SI and X-SI data (n=279) were included (558 SI-joints). Kappas and % agreement for all SL are presented on patient and SI-joint level (table). Despite high levels of agreement (e.g. 96.4% for ankylosis), the kappas for the comparison of all SL on MRI-SI and X-SI were rather low. This discrepancy is well known if abnormalities are infrequent: if there are no abnormalities agreement is high, yet positive findings are infrequently seen on both imaging modalities. Erosions were more frequently detected on MRI only; sclerosis and ankylosis more frequently on X-SI. MRI-SI-s+ was more frequent than mNY.
Conclusions In pts with recent onset axSpA, including a high number of pts without lesions, the agreement in detecting SL on MRI-SI and X-SI is poor. Overall, more SL are detected on MRI. More research on what findings on both modalities truly mean is required in order to determine their respective places in diagnosing and classifying pts.
Weber A&R 2010;62:3048-58
van der Linden A&R 1984;27:361–8
Disclosure of Interest None Declared
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.